top of page


Better care through better quality control: Why quality control of DNA libraries is crucial for real-time sequencing of clinical cultures

F B Wells(2,4,5) E Costa Conceição(2,4,5) A Dippenaar(1,3) M de Diego Fuertes(1,3) V Rennie(1,3) T Heupink(1,3) R Warren(2,4,5) A Van Rie(1,3)

1:University of Antwerp; 2:DST NRF Centre of Excellence for Biomedical Tuberculosis Research; 3:Institute of Global Health; 4:Biomedical Research Institute, Stellenbosch University; 5:South African Medical Research Council Centre for Tuberculosis Research

‘Real-time’ whole-genome sequencing (WGS) of Mycobacterium tuberculosis (Mtb) from clinical primary-MIGT culture (CPC) is a powerful tool for patient care. CPC library quality control (QC) parameters are not stated in manuals. There is no guidance on which genomic libraries can be successfully sequenced. We evaluated the genomic libraries QCs of 286 patient samples that were MGIT cultured and assessed for presence of Mtb. DNA was extracted using CTAB protocol and assesses for quantity and quality using Qubit and nanodrop. Libraries were prepared with Illumina DNA-Library Prep Kit with 0.23-20 ng/µl of total DNA input, using 10 PCR cycles. Library fragments were assessed with LabChip® and TapeStation. A library pool of 3-5 samples were sequenced on mid-output MiniSeq cartridge. The MAGMA-pipeline was used for WGS analysis. Of 103 patient baseline samples, 81 were Mtb-positive, 16 culture-negative and 6 Mtb-negative. Of the 81 Mtb-positive, 78 passed DNA QC. Of these 78 libraries, 25 were visually categorised as “excellent”(single peak, no tailing/shoulder), 32 as “good”(single peak, visible shoulders and/or tailing), 4 as “borderline”(single or multiple peaks with shoulders and/or tailing), 9 as “critical”(small peak, no shoulders and/or tailing) and 8 as “failed” (no peak, or peak below 350 or above 1200 bp). All 70 samples not classified as failed were successfully sequenced. A model was constructed using Amazon-Recognition, which had an F1-score of 0.857 for sequencing failures using a training dataset of 125 TapeStation images. Standardizing the CPC library fragment sizes range (350-1200 bp) and an automated visual-processing-tool can help streamline real-time WGS.

bottom of page